> > PEBS output Inte PT introduces some new MSRs (MSR_RELOAD_FIXED_CTRx)
> > for fixed function counters that use for autoload the preset value
> > after writing out a PEBS event.
> >
> > Introduce base MSRs address parameter to make this function can get
> > performance monitor counter structure by MSR_RELOAD_FIXED_CTRx
> registers.
> >
> > Signed-off-by: Luwei Kang <luwei.kang@xxxxxxxxx>
> > ---
> > arch/x86/kvm/pmu.h | 5 ++---
> > arch/x86/kvm/vmx/pmu_intel.c | 14 +++++++++-----
> > 2 files changed, 11 insertions(+), 8 deletions(-)
> >
> > diff --git a/arch/x86/kvm/pmu.h b/arch/x86/kvm/pmu.h index
> > 58265f7..c62a1ff 100644
> > --- a/arch/x86/kvm/pmu.h
> > +++ b/arch/x86/kvm/pmu.h
> > @@ -93,10 +93,9 @@ static inline struct kvm_pmc *get_gp_pmc(struct
> > kvm_pmu *pmu, u32 msr, }
> >
> > /* returns fixed PMC with the specified MSR */ -static inline struct
> > kvm_pmc *get_fixed_pmc(struct kvm_pmu *pmu, u32 msr)
> > +static inline struct kvm_pmc *get_fixed_pmc(struct kvm_pmu *pmu, u32
> msr,
> > + int
> > +base)
> > {
> > - int base = MSR_CORE_PERF_FIXED_CTR0;
> > -
> > if (msr >= base && msr < base + pmu->nr_arch_fixed_counters)
> > return &pmu->fixed_counters[msr - base];
>
> IIUC, these new MSRs aren't new fixed PMCs, but are values to be reloaded
> into the existing fixed PMCs when a PEBS event has been written. This change
> makes it look like you are introducing an additional set of fixed PMCs.
Yes, you are right. They are not new fixed counters.
Each fixed/general purpose counters have a "kvm_pmc" structure in KVM. We already have a function to get general purpose counter by event selectors and gp counters, as below:
pmc = get_gp_pmc(pmu, msr, MSR_IA32_PERFCTR0) //by gp counter
pmc = get_gp_pmc(pmu, msr, MSR_P6_EVNTSEL0) //by gp event selector
So I extended the " get_fixed_pmc " function to support get fixed counters by MSR_RELOAD_FIXED_CTRx. Actually, they are all get "kvm_pmc" structure by offset.
Thanks,
Luwei Kang
